- Last post
- 41 Responses
My favorite part about this is when it’s AI text, this forum loses its collective shit (in a bad way). But when it’s images, shit is again lost (but in a good way?)
So a computer making you feel like a creative without having to work to achieve the outcome you describe is good, but if a computer generates a story that you have to read, it’s bad?
Dall-e is nothing more than the most impressive hack you’ve never met. No original ideas, everything is a result of billions of images to train from as “inspiration”. Without all human effort before it, dall-e couldn’t exist.
Until this thing begins wielding a brush and creating physical works of art, it’s simply a replacement for all of us. Why hire a dipshit with an opinion when dall-e will give me jpg options to choose from in seconds?
Animations are coming next.
How long until dall-e can animate an entire screenplay complete with lighting?
It’s not like there isn’t a massive archive of movies and broadcast television to train it with.
Within 10 years, a tool like this will be able to compile complete cinematic experiences including dialogue soundtrack and arced storytelling.
As long as humans feed it what we like, it will only regurgitate what humans think they like.
Eventually, Disney will own this.
- So, a Marvel Movie generator?Nairn
- 40 years from now: generate new episodes of Star Trek TNG, new plot, but indistinguishable from the originaldrgs
- This scares me a littledrgs
- It really is nutsinv
- I think it’s closer to 7years than 40. around 2030.uan
- I have my doubts. It would need to generate sounds/voices, understand language, model human behavior etc, its much more than just an imagedrgs
- everything you mentioned is already being synthesized. The YT channel Two-minute papers shows countless techniques that would be implemented to accomplish this.imbecile
- All existing AI programs are specialized in one thing or another, image generators cannot render meaningful text:
- Language AI only do text etc
I haven't seen any attempts to bring it all togetherdrgs
- Are you stating that you don't believe a way to integrate these technologies together will ever exist? I believe differently.imbecile
- Who says I don't believe? I said in 40 yearsdrgs
- a pipeline of process, checks, and balances is all that is needed to produce content. Will a tool like this ever create anything original unprompted? noimbecile
- Will it create content based on whatever variables humans throw at it? Most definitelyimbecile
- You said you have your doubts which is universal for I don't believe, so to answer your question... you did.imbecile
- If you look at uan's post, right before drgs, you might think to lean back on being quite so absolutist in your interpretation of what drgs said.Nairn
- Btw, virtually all AI images have this "AI look": small defects in rendering which are immediately recognizable. Solve this first, then we'll talk about videodrgs
- I used midjourney for a project recently. Generated around 2k images, used parts from like 80. It def still requires art direction and PS skill at this pointscarabin
- I’m finding its real strength at the moment is in generating rapid “thumbnails” for larger hand-built pieces, and for low-res texturesscarabin
- you can listen to massive tons of a.i generated music.. no one gives a shit... will higly likely never take of in generalneverscared
- for people with puny design skills it must be good fun though.neverscared
- Current AI imaging i feel could be good for cranking out quick/cheap editorial “illustrations” for news articles and blog postsscarabin
- What crap and cheap tricks are passed as "AI" these days *rolling eyes*grafician
- This is the great equalizer. Imagine on 20 years time we can create entire movies, Hollywood quality without the cast or crew of thousands on a budget of..._niko
- ...just your time._niko
- that didn't take long... text by google imagen https://www.youtube.…imbecile
- easy fix it will be regulated.neverscared
Midjourney is quiet impressive too.
Is there some secret trick to getting into the betas for these things?
- Dall.e Mini is available but traffic limited. It is a simpler version of Dall.e 2. https://huggingface.…inv
- If you get the too much traffic message, just keep trying.inv
- appreciate the tip, I have however already been using mini, which is still amazing to play around with but the other stuff is next levelkalkal
- I'm waiting until it becomes "give up your email for spam"-servicedrgs
AI generated imagery is getting better astonishingly quickly. DALL-E to DALL-E 2 is a huge leap.
I can see tons of uses for this, not necessarily to replace illustrators or photographers completely just yet but I could see services like getty stock images becoming somewhat obsolete.
How long have you spent searching image libraries in the past for just what you're looking for? Relying on a huge pool of expensive imagery that getty have collected. Far more efficient to just tell an AI exactly what you want and have it generate something custom and unique for you. Some of the demos I've seen already allow you to specify a certain camera lens type and lighting conditions etc.
In regards to the comparison to human creatives...isn't it just doing what we have all done in some ways? We've spent a lifetime absorbing and internalising other content and then recalling it and rehashing at will in order to create something new (whether that recall is conscious or not). I don't have a problem with this development, it's incredible to see it progress so quickly.
I found this article interesting on the subject:
- yeah, i don't see it ever being an end solution for anything. i used it to its fullest potential in my workflow and 80% of my time was still spent in photoshopscarabin
- i spent the same amount of time as i normally would on stock images but like only $60 instead of $1000+. just wish they offered much, much higher resolutionscarabin
- just "feed" those huge stock photo libraries into DALL-E v3?grafician
- I was generating images last night on MidJourney and some of them had watermarks that were takes on the shutterstock watermark. lolmicrokorg
Cool. You got access to v2??
- fax me your CB handlefadein11
- You got invites?inv
- No invites... You can refer but chances are slim if you're not someone with a big following.Jrdntnnr
- I referred a friend with 600k subs on YT and they said he can get on in a few weeks. I referred an ex Pixar director and he got on in a few hours.Jrdntnnr
- The fake moon in the background is a giveaway I thinkdrgs
- Where can you generate ideas? Or is it still closed beta?NBQ00
- I have no ideadrgs
- this is hilariousmonospaced
- Can I ask, where are these being generated.. site?hydro74
- Search twitter +
- Or you mean